Categorical Analysis of Code Smell Detection Using Machine Learning Algorithms

In software development, a code smell is some surface indication that possibly corresponds to a deeper problem in the programme. Whether a piece of code has a code smell or not is very subjective and it varies mostly by the language and developer, but also the developmental technologies. The term code smell was popularised by Kent in the late 1990s. Since then several machine learning algorithms have been employed in order to detect code smells. Code smells have been categorised into several categories. With the lack of studies done in respect to smells and their relation to their respective classes, it is imperative to look into it for a deeper understanding of smells. In this paper, we have tried to analyse and study the detection accuracies of these smells categorically for unambiguous understanding of how smells work for efficient refactoring.